[clang] [clang][analyzer] Add taintedness to argv (PR #178054)

[Donát Nagy via cfe-commits]

================
@@ -1413,9 +1371,16 @@ For a more detailed description of configuration options, please see the
 
 **Configuration**
 
-* `Config`  Specifies the name of the YAML configuration file. The user can
+* ``Config`` Specifies the name of the YAML configuration file. The user can
   define their own taint sources and sinks.
 
+* The if the analyzer option ``assume-controlled-environment`` is set to ``false``,
+  it is assumed that the command line arguments and the environment
+  variables of the program are attacker controlled.
+  In particular, the ``argv`` argument of the ``main`` function and
----------------
NagyDonat wrote:

> As envp is not in the POSIX standard did not look important.

Support for `envp` is pretty widespread (e.g. [GCC supports it](https://www.gnu.org/software/c-intro-and-ref/manual/html_node/The-main-Function.html), [MSVC also supports it](https://learn.microsoft.com/en-us/cpp/cpp/main-function-command-line-args?view=msvc-170)) so I think we should recognize it and mark its contents as tainted. Also note that the C99 standard [explicitly allows implementations to accept more signatures for `main` than the usual two](https://en.cppreference.com/w/c/language/main_function.html).

>  The taintedness of the environment variables are handled when they are accessed through getenv() which is typically used to access them.

I agree that `getenv()` is probably more common (although I don't have concrete statistics), but we should still support `envp` if it is not too difficult (I assume that it can be handled the same way `argv` is handled).

https://github.com/llvm/llvm-project/pull/178054